On Mon, 17 Apr 2000, Laine Stump wrote:

>At 10:11 AM 4/17/00 +0100, Roger Brooks wrote:
>>However, please read the warning in the message from my original posting
>>anout PCI bus speeds other than 33 MHz (although I don't think you can
>>run a BP6 with any other PCI bus clock).
>
>I thought that the PCI bus clock was always either 1/3 or 1/2 of the FSB
>clock (depending on whether it's closer to 100Mhz or 66Mhz), meaning that
>any time you set the BIOS to a clock frequency other than 100 or 66, the
>PCI clock would be at something other than 33Mhz. For example, I usually
>run my BP6 FSB at 82Mhz (yielding 451Mhz from 366Mhz-rated parts), so my
>FSB would be 41Mhz (I think the switchover to 1/3-FSB happens at a higher
>speed). Note that running at this speed gives me problems with some PCI
>peripherals (in particular, all vintages of NetGear FA310-TX), a good
>indicator that the PCI clock is modified.

I'll rephrase that -- so far I can't get my system to overclock, hence I
can't get any PCI clock other than 33 MHz.  I have a pair of 400 MHz
Celerons.  At the time I bought the board (August 1999) there were rumours
that Intel might disable the SMP capability in the near future, and 400 MHz
chips gave the best price/performance ratio over here.  I'm not sure if
I have the speed-locked chips, but all attempts at overclocking have
resulted in the BIOS resetting to 300 MHz.

I'm not particularly interested in overclocking.  I got the board to follow
the SMP project, and my personal view is that overclocking when you're
running an experimental operating system is crazy, because if you hit a
problem you'll never be quite sure whether its hardware or software.

I would like to be able to overclock in order to get/test the HPT366 tuning
parameters for other PCI bus speeds (for the benefit of those who want to
risk it).  However, Highpoint Tech still haven't responded to my request
for documentation, so I can only get the tuning parameters for overclocked
PCI bus speeds if the BIOS sets them up correctly.

And if the BIOS (and the Win32 driver) don't DTRT on an overclocked system,
perhaps this explains why so many people seem to have trouble with the
HPT366.  My (non-overclocked) system seems to be very stable (~32 days
uptime so far), but I found that the effects of using the wrong HPT366
tuning parameters were catastrophic (silent write corruption, resulting
in a panic and just about the most FUBARed filesystems I've ever seen).


Roger

------------------------------------------------------------------------------
Roger Brooks (Systems Programmer),          |  Email: R.S.Brooks@liv.ac.uk
Computing Services Dept,                    |  Tel:   +44 151 794 4441
The University of Liverpool,                |  Fax:   +44 151 794 4442
PO Box 147, Liverpool L69 3BX, UK           | 
------------------------------------------------------------------------------